List your company

Jailbreak Attack on Gemma-7b five finetuned variants

66.2Average ASRPGP
-2.64815.22633.150.974Dec 14, 2025

Evaluation results

MethodLinks
PGPAttacker knowledge (Threat Model)=Fully black-box2025.1266.2
AutoDan (white)Attacker knowledge (Threat Model)=White-box2025.1263.4
Guiding-GCGAttacker knowledge (Threat Model)=Pretrain-to-finetune2025.1244.2
SCAV (adaptation)Attacker knowledge (Threat Model)=Pretrain-to-finetune2025.1241.8
TUJA (adaptation)Attacker knowledge (Threat Model)=Pretrain-to-finetune2025.1241.6
AutoDan (adaptation)Attacker knowledge (Threat Model)=Pretrain-to-finetune2025.1236.2
GCG (white)Attacker knowledge (Threat Model)=White-box2025.1236
PAIRAttacker knowledge (Threat Model)=Fully black-box2025.1234.4
SEAAttacker knowledge (Threat Model)=Pretrain-to-finetune2025.1225.6
DI-GCGAttacker knowledge (Threat Model)=Pretrain-to-finetune2025.1214
GCG (adaptation)Attacker knowledge (Threat Model)=Pretrain-to-finetune2025.1213.4
PIFAttacker knowledge (Threat Model)=Pretrain-to-finetune2025.1212
GCG EnsembleAttacker knowledge (Threat Model)=Fully black-box2025.129
DANAttacker knowledge (Threat Model)=Fully black-box2025.126.4
LSGM_LILA (adaptation)Attacker knowledge (Threat Model)=Pretrain-to-finetune2025.124.4
L4AAttacker knowledge (Threat Model)=Pretrain-to-finetune2025.120